Americans Warned Against Using AI For Voting Decisions

As Americans cast their votes, many have relied on artificial intelligence-powered chatbots to gather information on candidates and policies. According to Pew Research, around 50 percent of recently surveyed US adults have used an AI chatbot at some point, and 8 percent have used these tools ahead of the midterms.

However, the information provided by these chatbots is often unreliable, with much of it being out of date or sourced from partisan websites. Even supposedly nonpartisan chatbots, such as OpenAI's ChatGPT and Google's Gemini, have been found to provide inaccurate information when it comes to questions about the midterms.

Research by Forum AI has highlighted the shortcomings of these chatbots, with almost 40 percent of election answers containing at least one factual error. For example, Google's Gemini chatbot incorrectly stated the deadline to request a mail-in ballot in Pennsylvania, and Claude conflated a lawmaker with a former school superintendent of the same name.

The problem lies in the chatbots' reliance on an increasingly problematic pool of online information. According to NewsGuard, mainstream chatbots often cite 'pink slime' websites, which present as mainstream news websites but are actually funded by political or ideological groups. These sites have proliferated ahead of the midterms, and chatbots are citing them frequently.

OpenAI's ChatGPT was found to be the worst offender, citing pink slime sites in over 70 percent of searches. In contrast, xAI's Grok chatbot performed best, citing these sites in 30 percent of cases. The issue is not just with the chatbots themselves, but also with the companies behind them, which have overpromised on the capabilities of AI.

Nina Jankowicz, executive director of the American Sunlight Project, notes that Americans are not prepared to navigate an election in the age of consumer AI. She warns that executives and companies have overpromised on what AI is capable of, and that it often 'makes things up' when it does not know the answer.

The limitations of AI-powered chatbots highlight the need for voters to be cautious when using these tools to inform their decisions. With the potential for misinformation and inaccuracies, it is essential for voters to verify information through reputable sources before making their decisions. This article is based on reporting by WIRED AI.